Ion implantation is commonly used to dope semiconductor material. For example, ion implantation has been used to form source and drain regions, and to adjust the threshold voltages of MOS transistors.
There are several advantages of using ion implantation to dope semiconductors. For example, diffusion requires heating a wafer to high temperatures (in the range of 1000.degree.-1400.degree. C.), whereas ion implantation does not. High temperatures may cause crystal damage but in any event cause the further diffusion of dopants in the wafer thereby changing the sizes of the doped regions. If the transistor uses submicron geometries, such size changes can materially affect the characteristics of the transistor. Further, by using ion implantation a wafer can be doped through a thin oxide layer, and a larger variety of masks can be used than by using diffusion. Ion implantation also generally allows more precise control of doping depth and concentration.
A typical ion implantation machine, shown in FIG. 1, includes an ion source 100 that creates dopant ions to be implanted. Dopant elements used are generally the same as those used in diffusion (for example, As, P, Sb, and B). FIGS. 2a and 2b show an ion source 100 and an ionization arc chamber 101, in detail. A gas containing dopant atoms is released into chamber 101 which must be kept at a high vacuum level to prevent air molecules from being ionized and implanted into a semiconductor wafer. The dopant gas source contains molecules in which the dopant atom is combined with other atoms. Dopant gas sources generally include those used in diffusion such as flourine-based gases (e.g. PF.sub.5, AsF.sub.5, PF.sub.3). The dopant atoms must be separated from these other atoms in order to provide the ion beam used to bombard the semiconductor wafer.
To separate the dopant atoms out of these molecules, chamber 101 is provided at one end with a filament 102, a wire typically made of tungsten or tantalum. Filament 102 emits electrons when heated by the passage of electric current. The current follows a path through filament 102 as shown in FIGS. 2a and 2b. When filament 102 is heated to a certain temperature, electrons are "boiled off" filament 102, into chamber 101 in the direction of the arrows, where the dopant gas source is located. The electrons collide with the molecules in the dopant gas source, and separate these molecules into atoms by ionizing them. A repeller plate 103, at the other end of chamber 101, is charged to some positive voltage and accelerates the electrons for more effective collisions and thus a higher ionization rate. A typical repeller plate produces a 3%-5% higher ionization rate.
In addition to the dopant atoms, the ionized dopant gas source contains the other atoms that were combined with the dopant atoms in molecules. The ion beam which will be focussed on the semiconductor wafer must contain only the desired dopant atoms. Thus, a typical ion implantation machine is provided with a mass analyzer 200 for separating the dopant atoms from other atoms. Once separated, the dopant atoms are accelerated by a device such as an acceleration tube 300, and focused by a device such as magnetic lens 400, into an ion beam. This ion beam is directed in a controlled fashion by devices such as beam traps, beam gates and scanners 500, onto semiconductor wafers 600.
The electron cloud produced by filament 102 is not homogenous within chamber 101, despite the force provided by repeller plate 103. As the distance from filament 102 increases, the density of the electrons decreases producing an electron depletion zone in a region R which is the most distant region of chamber 101 from filament 102. If the density were homogenous, and there were no depletion zone, both the number and the effectiveness of collisions between electrons and dopant gas source molecules would increase thus enhancing the performance of the ion source.
Due to the large currents that flow through filament 102, filament 102 must be regularly replaced. Filament replacement is responsible for a large percentage of down-time of an ion implantation machine. Replacing a filament is an involved multi-step process requiring careful execution. First the pressure in the source chamber must be vented from high vacuum to atmosphere. Next, the ion source must be removed and the filament replaced. Then the ion source must be re-installed and the system pumped back to high vacuum. This entire process typically takes from 2-3 hours. If a machine is otherwise running efficiently, the filament will typically require replacement from 10-30 times per month, which accounts for 40-50% of all downtime. It would be desirable to decrease the amount of downtime due to filament replacement.